Deep Think Now Accessible in the Google Gemini App: Here’s How to Try It.

Google is joyfully announcing the rollout of Gemini 2.5 Deep Think within the Gemini app, extending its AI reasoning model to Google AI Ultra subscribers. Starting Friday, August 1, subscribers to this enriching AI experience, available for $250 per month, will gain access to Deep Think. This innovative model was first unveiled at Google I/O 2025, showcasing technology designed to tackle complex math and coding challenges.

Deep Think is recognized as Google’s most advanced model yet, offering enhanced reasoning capabilities and the interesting ability to address intricate problems through parallel thinking techniques. In their official announcement, Google expressed that Deep Think delivers “more detailed, creative, and thoughtful responses” to various prompts.

According to Google, Gemini 2.5 Deep Think outshines other AI models, as indicated by scores from Humanity’s Last Exam (HLE), a rigorous AI evaluation. The model achieved an impressive score of 34.8% on this challenging test, which covers questions across math, sciences, and the humanities. In comparison, xAI’s Grok 4 scored 25.4%, while OpenAI’s o3 achieved 20.3%. Additionally, Google’s model excelled against AI models from xAI and Anthropic on LiveCodeBench6, a test focusing on coding tasks.

Furthermore, Google is excited to release the official version of their Gemini 2.5 Deep Think model, which received a gold-medal standard at the International Mathematical Olympiad (IMO) earlier this July. This version will be provided to a select group of trusted testers, including mathematicians and academics, reflecting the impressive solutions that the president of IMO described as “astonishing in many respects.”